sys_mincore: stop locking potentially huge amount of user memory
the current implementation performs uvm_vslock() on the
user-specified amount of memory. it isn't safe in general.
some might even consider it as a security issue.
this commit fixes it by splitting the user-specified range into
small ones which a temporary kernel buffer can cover. it's ok to
report potentially stale values as the mincore() api is inherently
racy in the first place.
while we still ought to invent a proper "lock user memory for
direct i/o" infrastructue, in this particular use case, it isn't
necessary or appropriate.
zfs: fix case insensitive / utf-8 normalized file names (cont.)
this was intented to be a part of an earlier commit.
("zfs: fix case insensitive / utf-8 normalized file names")
for some reasons, it seems i unintentionally dropped this hunk
when porting the commit from git to cvs.
lintpkgsrc: minor error handling and doc fixes
If unable to unlink a file, report this. While here, also fix incorrect
substitutions in the man page generation.
kitty: update to 0.46.1
Package changes:
- upstream updated to use go 1.26.0
Upstream changes:
- smooth scrolling and momentum based scrolling for a natural,
smooth and kinetic scrolling experience
- drag kitty tabs around with the mouse to re-order them, move them
to another kitty OS Window or even detach them into their own OS
Window
- resize kitty windows (aka splits) with the mouse
- many bug fixes
PR bin/60099 Add more test cases
More test cases to test $* expansion in a variety of
situations (all using $* unquoted - and in contexts
where field splitting could happen).
Several of these new text cases assisted with debugging
the $* expansion code in sh, avoiding issues that otherwise
might not be seen as potential problems.
PR/60069 NetBSD non booting on H3 Allwinner
Disable EFI runtime support on arm as it doesn't play nicely with many
FWs where things go south after/during the call to SetVirtualAddressMap
Also seen by cp- on Odriod-XU4
The Coprocessor Protocol Violation trap is not present on the 68010
(the vector is "unassigned, reserved"), so don't install a handler
for it on that CPU.
Fix a brain-f*rt in the special mac68k "nofault" bus error handling.
The information we need to pass along is packaged up for us neatly
in the stack frame and arguments being passed to trap(), so use those
to extract the %a2 value and faulting address.
Issue raised and fix tested by nat@